| Metric | lstm | cnn | naive | xgboost |
|---|---|---|---|---|
| diff_of_means | 0.114 | 0.202 | -0.509 | -0.528 |
| ratio_of_sd | 0.986 | 1.001 | 0.997 | 0.947 |
| amplitude_ratio_of_means | 0.932 | 0.935 | 0.956 | 0.818 |
| maximum_error | 0.348 | 0.120 | 0.126 | 0.256 |
| ks_mean_on_coarse_res_with_extremes | 0.132 | 0.161 | 0.137 | 0.165 |
| qqplot_mae | 0.333 | 0.590 | 1.482 | 1.541 |
| acf_mae | 0.035 | 0.059 | 0.034 | 0.080 |
| extremogram_mae | 0.021 | 0.039 | 0.041 | 0.019 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | cnn | naive | xgboost |
|---|---|---|---|---|
| diff_of_means | 0.121 | 0.209 | -0.567 | -0.594 |
| ratio_of_sd | 0.982 | 0.995 | 0.967 | 0.917 |
| amplitude_ratio_of_means | 0.929 | 0.928 | 0.959 | 0.809 |
| maximum_error | 0.368 | 0.122 | 0.127 | 0.247 |
| ks_mean_on_coarse_res_with_extremes | 0.069 | 0.058 | 0.041 | 0.118 |
| qqplot_mae | 0.353 | 0.609 | 1.651 | 1.735 |
| acf_mae | 0.033 | 0.058 | 0.019 | 0.072 |
| extremogram_mae | 0.009 | 0.016 | 0.017 | 0.017 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | cnn | naive | xgboost |
|---|---|---|---|---|
| diff_of_means | 0.189 | 0.281 | -0.444 | -0.472 |
| ratio_of_sd | 0.979 | 0.995 | 0.995 | 0.947 |
| amplitude_ratio_of_means | 0.928 | 0.927 | 0.958 | 0.811 |
| maximum_error | 0.357 | 0.124 | 0.111 | 0.249 |
| ks_mean_on_coarse_res_with_extremes | 0.056 | 0.062 | 0.053 | 0.089 |
| qqplot_mae | 0.552 | 0.818 | 1.294 | 1.379 |
| acf_mae | 0.033 | 0.059 | 0.031 | 0.082 |
| extremogram_mae | 0.011 | 0.019 | 0.016 | 0.012 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | cnn | naive | xgboost |
|---|---|---|---|---|
| diff_of_means | 0.146 | 0.232 | -0.266 | -0.284 |
| ratio_of_sd | 0.999 | 1.016 | 1.019 | 0.965 |
| amplitude_ratio_of_means | 0.947 | 0.983 | 0.946 | 0.718 |
| maximum_error | 0.378 | 0.138 | 0.163 | 0.259 |
| ks_mean_on_coarse_res_with_extremes | 0.146 | 0.122 | 0.176 | 0.285 |
| qqplot_mae | 0.425 | 0.677 | 0.775 | 0.838 |
| acf_mae | 0.048 | 0.056 | 0.065 | 0.139 |
| extremogram_mae | 0.021 | 0.022 | 0.021 | 0.032 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | cnn | naive | xgboost |
|---|---|---|---|---|
| diff_of_means | 0.163 | 0.253 | -0.319 | -0.337 |
| ratio_of_sd | 0.978 | 0.987 | 1.153 | 1.094 |
| amplitude_ratio_of_means | 0.917 | 0.922 | 0.962 | 0.922 |
| maximum_error | 0.383 | 0.136 | 0.112 | 0.216 |
| ks_mean_on_coarse_res_with_extremes | 0.195 | 0.189 | 0.304 | 0.294 |
| qqplot_mae | 0.504 | 0.737 | 1.031 | 0.983 |
| acf_mae | 0.043 | 0.062 | 0.086 | 0.097 |
| extremogram_mae | 0.029 | 0.039 | 0.100 | 0.020 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | naive | xgboost | lstm | cnn |
|---|---|---|---|---|
| diff_of_means | -0.134 | -0.170 | 0.177 | 0.263 |
| ratio_of_sd | 1.109 | 1.061 | 0.978 | 0.989 |
| amplitude_ratio_of_means | 0.961 | 0.884 | 0.925 | 0.935 |
| maximum_error | 0.114 | 0.224 | 0.365 | 0.131 |
| ks_mean_on_coarse_res_with_extremes | 0.287 | 0.299 | 0.230 | 0.243 |
| qqplot_mae | 0.631 | 0.541 | 0.538 | 0.766 |
| acf_mae | 0.068 | 0.096 | 0.032 | 0.050 |
| extremogram_mae | 0.034 | 0.025 | 0.016 | 0.025 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | naive | xgboost | cnn |
|---|---|---|---|---|
| diff_of_means | 0.146 | -0.165 | -0.198 | 0.236 |
| ratio_of_sd | 0.988 | 1.155 | 1.099 | 0.999 |
| amplitude_ratio_of_means | 0.922 | 0.963 | 0.901 | 0.925 |
| maximum_error | 0.381 | 0.108 | 0.215 | 0.128 |
| ks_mean_on_coarse_res_with_extremes | 0.203 | 0.276 | 0.297 | 0.230 |
| qqplot_mae | 0.431 | 0.878 | 0.724 | 0.688 |
| acf_mae | 0.043 | 0.085 | 0.106 | 0.063 |
| extremogram_mae | 0.012 | 0.053 | 0.020 | 0.023 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97